A lingering pandemic. Ongoing economic uncertainty. Geopolitical unrest.

As organizations examine budgets throughout the year to ensure they’re on track, global inflation, higher prices, and spiking energy costs are still impacting businesses. Everyone is looking for cost savings as budgets have tightened. 

When it comes to the IT budget, cloud infrastructure — key to powering the modern enterprise — is a focal point. It is also a hefty portion of the IT budget. In fact, companies’ average spending on cloud infrastructure in 2022 was 41% of their IT budget, according to Gartner. Gartner Vice President Analyst Sid Nag observes that today’s “inflationary pressures and macroeconomic conditions are having a push and pull effect on cloud spending.”  Cloud computing, he writes, “will continue to be a bastion of safety and innovation, supporting growth during uncertain times due to its agile, elastic, and scalable nature.”

How can organizations spend the right amount (for them) on cloud operations? How can leaders ensure they have needed cloud infrastructure despite shrinking IT budgets? How can they optimize their cloud spending by re-organizing workloads and trimming waste? Organizations can stay one step ahead of cloud concerns by reducing their TCO (total cost of ownership) in public, private, and hybrid multi-cloud environments. As you solidify your IT budget (and prepare to roll out new projects the rest of the year), here are three steps to cut your 2023 cloud spending up to two-thirds.

1. Double Down on Public Cloud — And Then Focus On Optimizing Use and Costs 

Let me begin by saying there is no reason why your organization shouldn’t be using the public cloud. You can begin by examining what workloads you are currently running in the cloud (public or private), the cost of doing so, and why things are configured that way. Tap one of the many cloud cost calculator tools available online to map out different scenarios for running a given workload on private vs. public cloud. Based on the results of your cost analysis, you can make an informed, data-driven decision about whether you should use a public cloud or build or migrate to a cost-effective private cloud (if you need to run certain workloads on-premises via hardware that is physically located on the premises). There is also the hybrid multi-cloud architecture (more on this in a moment).

If you choose to pursue a public cloud deployment, you can refer back to your cloud pricing calculator to estimate spending for the workload and select a public cloud provider (including Amazon Web Services, Google Cloud Platform, and Microsoft Azure). First, choose the right region to deploy workloads. An availability region is the geographic location of the cloud data center. Different regions offer different service qualities related to latency, pricing, resource quotas, etc. You should also factor in the regulatory requirements of a region as well as extras such as network transfer costs and storage costs in addition to the list price of computing.

Next, think about grouping resources that have the same lifecycle. Most clouds allow you to group resources in a way that models the lifecycle of the underlying resources. In other words, you can create, use, and destroy resources as a group, meaning you are less likely to end up with a stray IP address that will cost you money once you are finished with it. Also, consider containerization of workloads, as containers can be more efficient for running workloads versus a VM (virtual machine) workload. You will have other overhead in running K8 infrastructures… but depending on how you are running containers on public clouds, they can be a cost-effective measure.

Look for native features within your public cloud of choice to pay less for the same number of resources. There are committed/sustained use discounts, annual universal credits, pay-as-you-go with committed use, and reserved instance tiers and use models. There is often a trade-off between flexibility and higher discount levels in the public cloud, so think about your workloads and use case(s). While a reserved instance is one you reserve (or rent) for a fixed duration, a spot instance uses the spare (unused) capacity of the public cloud (often available for less than the on-demand price). Spot instances, which allow you to save up to 90% in cloud costs, are ideal for interruptible/stateless workloads.

Once you have deployed your workloads, you should continuously monitor everything to identify areas for potential discounts. Cloud advisors, cost analysis tools, and budget notifications can help with this. You may need to enact auto shutdowns, scheduled suspensions, and identify inactive storage. Also, you may rightsize, autoscale, and/or move some cloud storage to a cooler tier with a lower storage cost.

2. Evaluate If/When/Where Cost-Effective Private Clouds May Be Best

If we look at capital expenditures (CapEx) vs. operating expenses (OpEx), a public cloud will have very different costs than a private cloud. CapEx refers to costs organizations pay upfront (one-time costs). Public clouds have low CapEx; just attach a credit card to a billing system, and you are set. Meanwhile, OpEx may quickly spike depending on workload size and scale. Private clouds, however, may involve a large CapEx with hardware, renting data center space, hiring a dedicated person to operate the private cloud, etc. However, OpEx is low and remains flat with the fixed-capacity private cloud, regardless of the workload. TCO (total cost of ownership) is calculated by adding CapEx and OpEx.

A private cloud consists of hypervisors running virtualized or containerized workloads. For a private cloud, the majority of your costs will involve internal operations and maintenance, software licenses, and hardware/facilities. Optimize and lower these costs so that you are better able to tap into R+D efforts — and innovate as a company. At that point, you will have a cost-effective private cloud.

3. Think About When a Hybrid Multi-Cloud Approach Makes Economic Sense

When you examine public vs. private cloud TCO, you may encounter scenarios in which a hybrid multi-cloud approach is best. In other words, you run different workloads on different types of clouds in a combined architecture to achieve cost optimization. In fact, 82% of IT leaders have already adopted a hybrid cloud model, according to Cisco’s 2022 Global Hybrid Cloud Trends report.

While a small-scale internal CRM system would be best suited for a public cloud, a medium-scale online banking system could be ideal for a hybrid multi-cloud architecture. In this setup, you run the majority of workloads in a cost-effective private cloud with additional workloads outsourced to a public cloud. If we consider a large-scale scenario such as a video streaming system, a hybrid multi-cloud architecture is ideal. This system includes a data warehouse, a data lake, a video transcoding engine, and more. Aside from the substantial resources required to power the system, we will have daily spikes in which the system demands a higher traffic load than usual. Here, we could use highly scalable public cloud resources during heavy load periods.

 Also, consider using open-source projects, tools, and operating systems (such as Ubuntu) to optimize cloud costs. This way, you avoid hidden software licenses, subscriptions, and vendor lock-in. You can combine independent open-source projects to create a functional private cloud environment, for example. You may also consider outsourcing your private cloud to a managed service provider (MSP) to avoid the cost of hiring and training an operations team.

The above guidelines should help you examine, analyze, and optimize your organization’s budget. And while it’s impossible to predict the future, controlling the budget variables you can will put you and your organization in a proactive position rather than a reactive one.